NVIDIA Smashes Visual Compute Density Record With New GPU Server

Discussion in 'Industry News' started by Tim S, 10 Aug 2007.

  1. Tim S

    Tim S New Member

    Joined:
    8 Nov 2001
    Posts:
    18,879
    Likes Received:
    76
    London, UK - August 8, 2007— NVIDIA Corporation (Nasdaq: NVDA) continues to smash through visual computing performance barriers with the introduction of NVIDIA Quadro® Plex Visual Computing System (VCS) Model S4, a graphics server featuring a record number of graphics processing units (GPUs) in a standard 1U server form factor. The new server contains more Quadro graphics processing power in less space than has ever been achieved before, enabling interactive visualisation of extremely large-scale data sets by multiple users with a minimal footprint.

    Designed for data-centric applications such as off-line rendering, remote graphics applications and embedded visual computing, the Quadro Plex Model S4 server takes visual compute density to new extremes by incorporating four NVIDIA Quadro® FX 5600 professional GPUs in an industry-standard 1U rack configuration; dynamically allocating compute, geometry, shading and pixel processing power for optimised GPU performance. GPU computing for visualisation solves programming functionality available in the NVIDIA CUDA™ tool suite.

    The Quadro Plex Model S4 is designed to serve high-quality interactive visualisation of large models to multiple users. Its 6 GB frame buffer (1.5 GB per GPU) can process large textures with full-screen antialiasing while Shader Model 4.0 vertex and pixel programmability delivers ultra-realistic effects for OpenGL and DirectX 10 applications. The server’s accelerated 3D texture performance provides greater interactivity for the visualisation of large volumetric datasets often required in energy and scientific research.

    The Quadro Plex Series
    Until now, NVIDIA’s Quadro Plex product line has consisted of desk side units that can also be mounted side-by-side in a 3U rack space. The Model S4 server complements the rest of the product line by offering extreme compute density in 1U housing for solving the largest visualization and high-performance computing problems.

    The Quadro Plex VCS Model II is intended primarily for powering large arrays of monitors. Featuring four Quadro FX 4500 GPUs with a 512 MB frame buffer per GPU, the Model II can power up to eight frame-synched, dual-link DVI displays.

    The Quadro Plex Model III adds high-definition SDI output by replacing the GPUs in the Model II with two Quadro FX 5500s with a 1 GB frame buffer per GPU and dual HD SDI graphics-to-video output boards, powering two dual-link DVI displays plus two HD SDI outputs.

    The Quadro Plex Model IV is intended primarily for processing and visualisation of large data sets desk side, in a workbench environment. It features two Quadro FX 5600 GPUs with a 1.5 GB frame buffer per GPU for processing of larger textures and can power up to four frame-synched dual-link DVI displays.

    The Quadro Plex Model S4 is an industry standard form factor optimised for large-scale server deployments. Featuring four Quadro GPUs in a high-density 1U chassis for high-performance remote graphics and offline rendering applications, the Quadro Plex Model S4 offers system monitoring, thermal control and fault notification for efficient integration into data centers.

    NVIDIA Quadro Plex VCS Models I, II, III, and IV are currently available through NVIDIA and distribution partners worldwide, including PNY Technologies, Leadtek, and Elsa. The Quadro Plex VCS Model S4 will be available in the fourth quarter of 2007. Visit www.nvidia.com for Quadro Plex system configurations and additional product details.

    NVIDIA Corporation
    NVIDIA Corporation is the worldwide leader in programmable graphics processor technologies. The company creates innovative, industry-changing products for computing, consumer electronics, and mobile devices. NVIDIA is headquartered in Santa Clara, CA and has offices throughout Asia, Europe, and the Americas. For more information, visit www.nvidia.com.
     
  2. wewe

    wewe New Member

    Joined:
    16 Feb 2007
    Posts:
    119
    Likes Received:
    0
    Holy crap... now to find someone to make use of this technology..
     
  3. The_Beast

    The_Beast I like wood ಠ_ಠ

    Joined:
    21 Apr 2007
    Posts:
    7,379
    Likes Received:
    164
    Correct me if I'm wrong but it's stuff like this used for AutoCAD, Solidworks, Inventor, Photoshop.....
     
  4. identikit

    identikit Active Member

    Joined:
    5 Jun 2004
    Posts:
    1,320
    Likes Received:
    16
    True. Basically it unloads all the rendering to the GPUs.
     
  5. The_Beast

    The_Beast I like wood ಠ_ಠ

    Joined:
    21 Apr 2007
    Posts:
    7,379
    Likes Received:
    164
    I was pretty sure but wasn't positive, so thanks
     
  6. Sam0r

    Sam0r It's been a while

    Joined:
    24 Jun 2005
    Posts:
    1,900
    Likes Received:
    1
    Ah.. but will it play Crysis?
     
  7. The_Beast

    The_Beast I like wood ಠ_ಠ

    Joined:
    21 Apr 2007
    Posts:
    7,379
    Likes Received:
    164
    do you think a FX5600 could play Crysis??? no it can't
     
  8. beamifrend

    beamifrend New Member

    Joined:
    23 Aug 2007
    Posts:
    10
    Likes Received:
    0
    I use Maya with frames taking 3-4 minutes to render on a Xeon. A one minute animation at 24fps takes 3 days to render! This baby would do it in an hour!
     
  9. Icy EyeG

    Icy EyeG Controlled by Eyebrow Powers™

    Joined:
    23 Jul 2007
    Posts:
    516
    Likes Received:
    3
    A Quadro FX5600 is based on the 8800 series, not on the Geforce FX5600 :p

    EDIT: I can't see the difference between this one and the Tesla S870 GPU Computing Server, as the Quadro Plex Model S4 doesn't have any display outputs, just like the Tesla.
     
    Last edited: 23 Aug 2007

Share This Page